Piezo1, a mechanically activated ion channel,is required for vascular development in miceSanjeev S. Ranadea,b,1, Zhaozhu Qiua,b,c,1, Seung-Hyun Wooa,b, Sung Sik Hurd,e, Swetha E. Murthya,b,Stuart M. Cahalana,b, Jie Xua,b,c, Jayanti Mathurc, Michael Bandella,b,c, Bertrand Costea,b,2, Yi-Shuan J. Lid,e,Shu Chiend,e,3, and Ardem Patapoutiana,b,3

aHoward Hughes Medical Institute and bDepartment of Molecular and Cellular Neuroscience, The Scripps Research Institute, La Jolla, CA 92037; cGenomicsInstitute of the Novartis Research Foundation, San Diego, CA 92121; and dDepartment of Bioengineering and eInstitute of Engineering in Medicine, Universityof California, San Diego, La Jolla, CA 92032

Contributed by Shu Chien, May 21, 2014 (sent for review May 09, 2014)

Mechanosensation is perhaps the last sensory modality not un-derstood at the molecular level. Ion channels that sense mechan-ical force are postulated to play critical roles in a variety ofbiological processes including sensing touch/pain (somatosensa-tion), sound (hearing), and shear stress (cardiovascular physiol-ogy); however, the identity of these ion channels has remainedelusive. We previously identified Piezo1 and Piezo2 as mechan-ically activated cation channels that are expressed in many mecha-nosensitive cell types. Here, we show that Piezo1 is expressedin endothelial cells of developing blood vessels in mice. Piezo1-deficient embryos die at midgestation with defects in vascularremodeling, a process critically influenced by blood flow. Wedemonstrate that Piezo1 is activated by shear stress, the majortype of mechanical force experienced by endothelial cells in re-sponse to blood flow. Furthermore, loss of Piezo1 in endothelialcells leads to deficits in stress fiber and cellular orientation inresponse to shear stress, linking Piezo1 mechanotransductionto regulation of cell morphology. These findings highlight an es-sential role of mammalian Piezo1 in vascular development duringembryonic development.

Mechanotransduction, the conversion of physical forces intoa biochemical response, is an essential signaling mecha-

nism used by all organisms (1). Diverse physiological processesincluding hearing, touch, and blood pressure regulation requirethe ability of cells to sense and respond to mechanical stimuli (2–4). Impairment in mechanotransduction can lead to a wide rangeof pathologies including deafness and atherosclerosis (5, 6). Thecellular response to mechanical force requires the coordinatedaction of multiple proteins; however, the identity of mechano-transducers in mammals remains poorly understood (7).Classical studies from auditory sensory cells first implicated ion

channels to be inherently mechanically activated (8). Since then,stretch-activated cation channels have been recorded from varioustissues (9), but their identity has remained unknown (10, 11). Foran ion channel to be considered a physiologically relevant trans-ducer of mechanical stimuli, several criteria must be met (12). Theion channel must be expressed in the appropriate mechanosensi-tive cells, its expression should be required for mechano-transduction in an in vivo setting, and ideally the expression of thechannel in a naive cell must render that cell mechanosensitive.Progress toward identification of mechanically activated (MA)

ion channels has been limited to a subset of ion channel families.In the invertebrates Caenorhabditis elegans and Drosophila mel-anogaster, members of the DEG/ENaC family and the TRP ionchannel NOMPC have been conclusively shown to act as mecha-notransduction ion channels (13–16). Mammals, however, do nothave an ortholog of NOMPC, and the physiological relevance ofthe mammalian ENaC orthologs in mechanotransduction remainsunclear. Many other TRP channels including TRPPs, TRPA1,and TRPV4 have been implicated in sensing mechanical stimuli;however, none has been shown to be required in vivo forsensing mechanical forces, nor to be necessary and sufficient

for mechanotransduction (12). The two-pore potassium channelsTREK-1, TREK-2, and TRAAK are mechanically activated inheterologous expression systems, and TREK-1 plays a role inmodulating the mechanical sensitivity of dorsal root ganglianeurons (17). Overall, the identification of relatively few candi-date genes has been a major impediment to studying mechano-transduction processes in vivo.We recently identified Piezo proteins as pore-forming sub-

units of an evolutionarily conserved MA cation channel family(18). Mouse Piezo1 forms a homo-multimeric complex of ∼1.2mDa, with each monomer containing over 35 transmembranedomains (19). Piezo1 is necessary for mechanically activatedcurrents in the Neuro2A cell line and is sufficient for confer-ring stretch-activated currents in heterologous cell expression.Purified mouse Piezo1 retains channel activity when recon-stituted without accessory subunits into artificial lipid bilayers.The single Piezo member in Drosophila plays a role in sensingnoxious mechanical stimuli (20), and mouse Piezo2 has beenshown to be critically required for mechanotransduction inmammalian Merkel cells (21–23). Gain-of-function PIEZO1mutations have recently been shown to cause hereditary xero-cytosis in humans (24–26); however, it is still unclear whetherthe mammalian Piezo1 plays a role in mechanotransduction

Significance

Ion channels that are activated by mechanical force have beenimplicated in numerous physiological systems. In mammals, theidentity of these channels remains poorly understood. We re-cently described Piezos as evolutionarily conserved mechan-ically activated ion channels and showed that Piezo2 is requiredfor activation of touch receptors in the skin. Here we show thatPiezo1 is a critical component of endothelial cell mechano-transduction and is required for embryonic development. Piezo1is expressed in embryonic endothelial cells and is activated byfluid shear stress. Loss of Piezo1 affects the ability of endothelialcells to alter their alignment when subjected to shear stress.These results suggest a potential role for Piezo1 in mechano-transduction in adult cardiovascular function and disease.

Author contributions: S.S.R., Z.Q., S.S.H., Y.-S.J.L., S.C., and A.P. designed research; S.S.R.,Z.Q., S.-H.W., S.S.H., S.E.M., S.M.C., J.X., J.M., M.B., and B.C. performed research; S.S.R.,Z.Q., S.-H.W., S.S.H., Y.-S.J.L., and S.C. analyzed data; and S.S.R., Z.Q., S.S.H., S.C., and A.P.wrote the paper.

The authors declare no conflict of interest.1S.S.R. and Z.Q. contributed equally to this work.2Present address: Ion Channels and Sensory Transduction Group, Aix Marseille Université,Centre National de la Recherche Scientifique, Centre de Recherche en Neurobiologieet Neurophysiologie de Marseille, Unite Mixte de Recherche 7286, 13344 Marseille,France.

3To whom correspondence may be addressed. E-mail: shuchien@ucsd.edu or ardem@scripps.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1409233111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1409233111 PNAS | July 15, 2014 | vol. 111 | no. 28 | 10347–10352

PHYS

IOLO

GY

Dow

nloa

ded

by g

uest

on

July

25,

202

0

in vivo. Here we used a genetic strategy to ablate the function ofPiezo1 in vivo and show that Piezo1 is a critical component ofendothelial cell mechanotransduction.

ResultsPiezo1gt/gt Mice Are Embryonic Lethal. To characterize the physio-logical function of Piezo1, we used a gene trap transgenic linein which a promoter-less β-geo (β-gal and neomycin phospho-transferase) cassette had been randomly integrated in the intronbetween exons 18 and 19 of the mouse Piezo1 locus (Fig. 1A). Thegene trap cassette contains a splice acceptor that co-opts the en-dogenous splicing of the Piezo1 transcript, leading to an in-framefusion protein containing the first 823 residues of Piezo1 andβ-geo, a predicted nonfunctional protein. This gene trap allele,Piezo1gt, also provides two reporters of Piezo1 expression: β-galfused to Piezo1 and human placental alkaline phosphatase (PLAP)expressed through an internal ribosome entry site (IRES) (27).We confirmed proper splicing of the Piezo1gt allele by per-

forming RT-PCR with primers targeting the Piezo1 transcriptsurrounding the integration site and the start of the gene trapcassette (Fig. S1A). We then cloned and expressed the entirePiezo1gt transcript in human embryonic kidney (HEK293T) cells(Fig. S1B) and found no cellular toxicity in three different assays(Fig. S1C). Furthermore, we showed that an N-terminal frag-ment of Piezo1, Piezo1 (1–823), fused to eGFP did not lead toany MA currents (Fig. S1D). These results indicate that Piezo1gt

would lead to loss of Piezo1 ion channel activity while serving asa reporter of endogenous Piezo1 expression.Piezo1+/gt mice were viable and fertile, but intercrossing Piezo1+/gt

mice generated no viable homozygous offspring (Fig. 1B).Quantitative PCR (qPCR) analysis using a probe downstream ofthe gene trap integration site confirmed loss of wild-type Piezo1transcript (Fig. 1C). Piezo1gt/gt embryos seemed normal at em-bryonic day (E) 9, showing a regular heartbeat and blood flow incentral vessels, and, at E9.5 major blood vessels such as thedorsal aorta and cardinal veins were present in Piezo1gt/gt em-bryos, suggesting that the initial stages of vasculogenesis occurrednormally in the absence of Piezo1 function (Fig. S2A). However,starting from E9.5, mutant embryos were growth-retarded (Fig. 1D and E), and many exhibited pericardial effusion at E10.5, in-dicative of poor cardiovascular function (28). A majority ofPiezo1gt/gt embryos died at midgestation before E14.5 (Fig. 1B).

Piezo1 Expression in Embryonic Endothelial Cells. Embryonic le-thality beginning at E9.5 coincides with the development of thecardiovascular system, whose function is required for the survivalof the embryo (28). We investigated where Piezo1 is expressed atthis time. Piezo1 in situ hybridization probes, an antibody thatwe had previously generated, as well as commercially availableantibodies were not sensitive enough to detect expression of en-dogenous Piezo1 (18, 29). We therefore used the Piezo1 pro-moter-driven β-gal and PLAP reporter genes to analyze theexpression pattern of Piezo1. Weak β-gal activity was initiallydetected at E9 (Fig. S2B), when regular heartbeats and blood floware first established (29). Between E9 and E10.5, β-gal activityincreased and was restricted to the developing cardiovascularsystem, in particular to the vascular endothelium and the innerendocardium surrounding the heart lumen (Fig. 2A). Staining ofPLAP, a GPI-linked cell surface marker, and platelet/endothelialcell adhesion molecule 1 (PECAM1), an endothelial marker, alsoconfirmed Piezo1 expression in the vascular endothelium (Fig.2B). In the heart, the highest level of X-gal staining at E11.5 wasobserved in the endothelium lining ventricular outflow tract andatrioventricular canal (Fig. 2C), areas predicted to experiencestrong fluid shear forces generated by blood flow (30, 31).To provide an even more sensitive method to detect Piezo1

expression, we designed a gene-targeting strategy that would leadto a C-terminus fusion protein of Piezo1 with the fluorescentreporter tdTomato (Fig. S3A). Germ-line transmission and suc-cessful deletion of the neomycin cassette led to mice thatexpressed the fluorescent reporter allele Piezo1P1-tdT (Fig. S3B).RT-PCR analysis of Piezo1+/P1-tdT mice confirmed the presenceof a fusion allele with the expected size (Fig. S3C). Western blotanalysis confirmed the presence of a Piezo1 fusion protein inPiezo1P1-tdT/P1-tdT mice (Fig. S3D). These mice were viable andexpressed Piezo1 in levels similar to Piezo1+/+ controls (Fig. S3E).We then used flow cytometry to detect the epifluorescence of

tdTomato with single-cell resolution. We show that PECAM1+

endothelial cells in the embryonic heart and yolk sac of midgestationPiezo1+/P1-tdT embryos display a prominent shift in fluorescencepeak compared with Piezo1+/+ littermates, indicating expressionof the Piezo1–tdTomato fusion allele (Fig. 2D). We further an-alyzed cells within the yolk sac that were PECAM1− as a con-trol (Fig. 2E). The majority of yolk sac-derived nonendothelialcells did not show fluorescence above background, confirmingour lacZ staining results that Piezo1 is preferentially expressedin endothelial cells. Interestingly, expression of Piezo1 persistsin endothelial cells in adult heart, lung, and aorta, as well as inthe human umbilical vein endothelial cells (HUVECs) (Fig. S4A and B).

Vascular Impairments in Piezo1gt/gt Embryos. Blood flow in mouseembryos begins at E8, and hemodynamic forces from blood flowhave been shown to be a critical factor to the development of the

Fig. 1. Piezo1-deficient embryos die at midgestation. (A) Diagram of thePiezo1 genomic locus, gene trap cassette, and the products of Piezo1 genetrap allele (Piezogt ). The exons (black rectangles), β-geo (a fusion betweenβ-gal and neomycin phosphotransferase, blue rectangle), and PLAP (purplerectangle) are indicated. (B) Percentage of viable Piezo1gt/gt mice at differ-ent ages derived from Piezo1gt/+ intercrosses. The number of embryos andmice genotyped is shown in the bottom. (C) Relative mRNA expression ofPiezo1 in mouse embryonic fibroblasts (MEFs) isolated from E10.5 Piezo1gt/gt

embryos compared with Piezo1+/+ MEFs. mRNA levels were determined byquantitative real-time PCR and normalized to glyceraldehyde 3-phosphatedehydrogenase (Gapdh). Bar represents the mean ± SEM (n = 3). (D) Grossappearance of Piezo1gt/gt embryos and their Piezo1+/+ littermates. Red arrowindicates pericardial effusion. (Scale bar: 1 mm.) (E) Size analysis of repre-sentative Piezo1+/+, Piezo1+/gt, and Piezo1gt/gt embryos at E9.5 and E10.5.Numbers of embryos usedwere Piezo1+/+, E9.5 (n = 10), E10.5 (n = 9); Piezo1+/gt,E9.5 (n = 20), E10.5 (n = 13); Piezo1gt/gt, E9.5 (n = 9), E10.5 (n = 8). Bars representthe mean ± SEM *P < 0.05, ***P < 0.001, unpaired two-tailed t test.

10348 | www.pnas.org/cgi/doi/10.1073/pnas.1409233111 Ranade et al.

Dow

nloa

ded

by g

uest

on

July

25,

202

0

refined vascular network (32, 33). The initial stage of vascularremodeling relies entirely on endothelial cells because muralcells consisting of smooth muscle cells and pericytes present inmature vessels are not recruited until after the primary vesselstructure has been established (34). Having shown that Piezo1 isexpressed in embryonic endothelial cells, we characterized themorphology of the developing vascular system of Piezogt/gt mice.We performed H&E staining and PECAM1 immunostaining

to observe vessel morphology in the yolk sac and the embryoproper from E8.5 to E10.5, when vascular remodeling and net-work formation occur. PECAM1 staining of E8.5 yolk sac showsthat the organization of the early vascular plexus is significantlyaltered in Piezo1gt/gt yolk sacs (Fig. 3 A and B). The layer of en-dothelial cells that cover the mesodermal cells of the yolk sac iswider in Piezo1gt/gt embryos, and the vasculature is delayed in itsorganization of major vessels. By E9.5, the yolk sac vasculature isless defined and the number of major vessels is fewer relative tolittermate controls (Fig. S5A). H&E staining of a cross-section ofthe E9.5 yolk sac shows that wild-type yolk sacs have a wide

distribution of small- and large-diameter vessels, whereasPiezo1gt/gt yolk sacs lack major vessels (Fig. 3 C and D). Vascularremodeling in the Piezo1gt/gt embryo proper reveals a similarlack of vessel organization at E9.5 and a significant decrease inthe number of major vessels at E10.5 compared with Piezo1+/+

littermate controls (Fig. 5SB). Histological analysis of the heartat E8.5 did not reveal obvious morphological differences be-tween Piezo1gt/gt embryos and littermate controls (Fig. 5SC).These data indicate a role for Piezo1 in the remodeling of theembryonic vasculature and suggest that a loss of functionalPiezo1 leads to an impairment of mechanotransduction pro-cesses in vivo.

Piezo1 Is Activated by Shear Stress. We have previously shown thatPiezo1 is activated by two types of distinct mechanical stim-ulations: by the indentation of a cell with a glass probe and bystretching the cell membrane (18). Although membrane stretchcan affect endothelial cells, fluid shear stress represents the mostrelevant force experienced by endothelial cells (35, 36). Wetested shear stress activation of Piezo1 using electrophysiology.We used a perfusion tube and subjected cells to laminar flowshear stress by applying 600-ms pulses of solution flow directedon to the recording cell (Fig. 4A). Whole-cell inward currents(Imax: −533 ± 64 pA) could be recorded from Piezo1-IRES-eGFP transfected HEK293T cells when stimulated by shearstress in the intensity range of 52 ± 3–64 ± 2 dynes/cm2, but notfrom IRES-eGFP transfected cells (Imax: −8.7 ± 1.3 pA) even upto 70 dynes/cm2 (Fig. 4B). The pressure applied to the solutionwas correlated to shear stress by using the particle image veloc-imetry method to estimate the shear stress intensity at the cellsurface (37). For pressure pulses ranging from 10 mmHg to 60mmHg, the average bead velocity ranged from 6.9 ± 0.6 μm/msto 21.0 ± 1.3 μm/ms, and the estimated wall shear stress intensityranged from 23.3 ± 2.2 dynes/cm2 to 70.2 ± 4.2 dynes/cm2 (Fig.4C). The half-maximal activation intensity for Piezo1 was mea-sured to be 57 ± 3 dynes/cm2 (Fig. 4D). Similar to MA currents,shear stress-activated Piezo1 channel currents inactivate incontinued presence of stimulus, but with a slower time constant

Fig. 2. Piezo1 is expressed in embryonic vascular endothelial cells. (A) Rep-resentative images of X-gal–stained whole-mount Piezo1+/+ and Piezo+/gt

embryos at E9.5 and E10.5. (B) Representative image of E9.5 whole-mountPiezo1+/+ and Piezo1+/gt embryos stained for PLAP and PECAM1 and magni-fication image of embryonic head showing expression of PLAP and PECAM1from the same embryos. (C) Representative images of X-gal–stained E11.5whole-mount hearts and sections of Piezo+/gt embryos. (D) Representativehistograms of tdTomato fluorescence in PECAM1+ cells from Piezo1+/+ (grayshaded) or Piezo1+/P1-tdT (red line) embryos. E13.5 embryos and E11.5 embryoswere used for heart and yolk sac analysis, respectively, owing to technicalconsideration for tissue preparation. (E) Histograms of tdTomato fluorescencein PECAM1− cells from embryos in D. †Most PECAM1− cells from Piezo1+/tdT

yolk sac do not express Piezo1; ‡some PECAM1− cells in Piezo1+/tdT yolk sac alsoexpress Piezo1-tdTomato. Stainings in A–C were performed in at least fourembryos per genotype per time point. Histograms in D and E are represen-tative of at least two embryos per genotype. [Scale bars: 400 μm (A), 500 μmand 200 μm, respectively (B), and 200 μm (C).]

Fig. 3. Piezo1-deficient embryos display vascular remodeling defects. (A)Representative images of PECAM1-stained E8.5 Piezo1+/+ and Piezo1gt/gt

whole-mount yolk sacs. (B) Quantification of PECAM1+ area in E8.5 Piezo1+/+

and Piezo1gt/gt whole-mount yolk sac from n = 3 experiments. (C) Repre-sentative H&E staining of E9.5 yolk sac cross-sections from Piezo1+/+ andPiezo1gt/gt embryos. (D) Quantification of vessel thickness from H&E stainingof E9.5 yolk sac from n = 2 experiments. *P < 0.05, unpaired two-tailedStudent t test. [Scale bars: 100 μm (A) and 200 μm (C).]

Ranade et al. PNAS | July 15, 2014 | vol. 111 | no. 28 | 10349

PHYS

IOLO

GY

Dow

nloa

ded

by g

uest

on

July

25,

202

0

of 65.6 ± 9.0 ms (n = 8). Interestingly, it has been recently shownthat stretch-induced Piezo1 inactivation kinetics can vary dra-matically in different cell types, supporting the possibility thatlaminar shear stress (LS) can also modulate Piezo1 activity (38).Furthermore, physiologically relevant fluid shear stress fromblood flow comes in many forms, including laminar and dis-turbed flow. Disturbed flow during embryogenesis and in theadult is common in arterial branch points. As a rapidly adaptingion channel, disturbed flow would allow recovery of Piezo1 andcould activate the channel repeatedly. Overall, these data showthat Piezo1 can be activated by fluid shear stress, as well as bymembrane indentation and stretch.

Piezo1 Is Involved in Endothelial Cell Alignment Under Flow. Theability of endothelial cells to align in the direction of fluid flowis one of the best-studied cellular consequences of mechano-transduction (39). To determine whether loss of PIEZO1 leadsto defects in morphology of vascular endothelial cells and itsresponsiveness to shear flow, we compared the LS-inducedremodeling in HUVECs transfected with siRNA for PIEZO1and scrambled siRNA. HUVECs were subjected to shear stress(12 dynes/cm2) for 24 h and examined for changes in cellularmorphology. qPCR analysis revealed∼80% knockdown of PIEZO1transcript in PIEZO1 siRNA cells compared with scrambled

siRNA (Fig. S6A). The relatively long duration and the levelof shear stress were chosen based on previous studies on shear-induced remodeling of HUVECs (40).Stress fibers visualized by F-actin immunostaining and their

orientation angles show impaired alignment in PIEZO1 siRNAcells compared with scrambled siRNA (Fig. 5A). For HUVECstransfected with scrambled siRNA (Fig. 5 B and 5 C, Left),neither the stress fibers nor the cells showed any preferentialorientation under static condition (ST), but both showeddominant orientations parallel to the flow direction (0°) underLS, with 41 ± 0.58% of the stress fibers and 54 ± 0.38% of thecells having orientations within the range of ± 15°. Transfectionof HUVECs with siRNA for PIEZO1 (Fig. 5 B and C, Right)caused significant decreases in the LS-induced orientations ofstress fibers (25 ± 0.71%) and cells (31 ± 1.30%) in comparisonwith scrambled siRNA under LS. Morphological determinationof cell elongation in scrambled siRNA cells showed that LScaused an increase in aspect ratio to 3.3 ± 0.07 from the STvalue of 2.0 ± 0.04 (Fig. S6B, Left). Cells transfected withPIEZO1 siRNA showed a reduction in the LS-induced elon-gation, with an aspect ratio (2.6 ± 0.08) significantly lower thanthat with scrambled siRNA, whereas the ST value of 2.1 ± 0.04is not significantly different (Fig. S6B, Right). It is worth notingthat we observe a role for Piezo1 in alignment and elongation inresponse to lower forces of shear stress compared with forcesneeded to activate Piezo1 in studies of electrophysiology. Thisis not surprising, because the 24-h time frame of the experi-ments involving altered cell morphology allows accumulation oflow-level cation influx. These observations indicate thatknocking down PIEZO1 results in significant decreases in themechanosensitivity of the endothelial cells in terms of align-ment and elongation in response to LS. However, the de-pendence on PIEZO1 is partial, and it is likely that otherfactors also contribute to the shear stress-dependent alignmentand elongation.

DiscussionVascular development and cardiogenesis in the mouse embryois influenced by hemodynamic forces owing to blood flow. Forexample, vascular remodeling in the yolk sac is compromisedwhen cardiac function/output is reduced or absent (41). Surgicalligation of the left arch arteries in mouse embryos preventsnormal blood flow and leads to abnormal regression and anaccompanying enlargement of the right arteries (42). In addition,local hemodynamic shear stress forces imparted on vascularendothelial cells of developing blood vessels are necessary andsufficient to induce remodeling in the mouse yolk sac (43). Thesedata suggest that disruption of mechanotransduction in the car-diovascular system during the time when blood flow initiates aroundE8–E9 is expected to cause a defect in vascular development.The direct effect of hemodynamic force on endothelial cells

during vascular remodeling is not well understood. In responseto fluid shear stress in vitro, endothelial cells change their geneexpression profiles and can realign their cytoskeletal structure(35). Putative mechanotransducers have been identified fromcultured endothelial cells or in the mature vasculature, includingion channels (44, 45), the apical glycocalyx (46), cytoskeleton(47), primary cilia (48), and a protein complex comprisingPECAM1, vascular endothelial cadherin, and vascular endothe-lial growth factor receptor 2 (VEGFR2) (49). Ion channels suchas TREK-1 have been shown to be activated by stretch and shearstress (50). However, the identity of the shear stress sensor orsensors during embryogenesis has been unknown. Our data re-veal that Piezo1 is expressed in embryonic endothelial cells andthat Piezo1 deficiency leads to disruption of vascular de-velopment in the mouse embryo and yolk sac. In addition, ourresults provide evidence that Piezo1 is a sensor of shear stress,and that endothelial cell alignment in response to shear stress is

Fig. 4. Piezo1 is activated by fluid shear stress. (A) Pictorial representationof perfusion tube setup. Whole-cell currents were recorded from HEK293Tcells while the cell was stimulated by shear stress induced by directing pulsesof fluid flow through a perfusion tube placed relative to the cell. The flowrate is regulated by varying the pressure applied to the fluid inside the tube.(B) Representative traces of whole-cell currents recorded at −80 mV fromHEK293T cells expressing mPiezo1 (Left) or IRES-eGFP (Right) in response to600-ms pulses of increasing laminar shear stress. The stimulus trace abovethe current indicates the pressure pulse applied to the solution in the per-fusion tube (which was later back-calculated to the applied shear stressintensities) and corresponds to the whole-cell current recorded for thatstimulus. (Inset) Shear stress-induced average maximum whole-cell peakcurrents from mPiezo1 (n = 8) and IRES-EGFP (n = 6) transfected cells. Barsindicate mean ± SEM. (C ) Pressure–wall shear stress relationship. Particleimage velocimetry was used to estimate the wall shear stress intensity atthe indicated tube pressure. Each data point is the mean ± SEM fromvalues of five or six independent trials. The measurement was done for therange of pressure (0–60 mmHg) used to activate the cells during the ex-periment. (D) Wall shear stress to normalized peak current relationshipplotted for six individual cells expressing mPiezo1, depicting range of ac-tivation. For each cell, currents were normalized to the peak currentrecorded at maximum shear stress, before losing the seal. The stimulus–response curve was fit with a Boltzmann function to determine the half-maximal response.

10350 | www.pnas.org/cgi/doi/10.1073/pnas.1409233111 Ranade et al.

Dow

nloa

ded

by g

uest

on

July

25,

202

0

compromised in the absence of Piezo1. It is possible that in-complete endothelial cell alignment in response to shear stressduring remodeling of arteries leads to the vascular remodelingdefects observed in Piezo1-deficient embryos. Alternatively,Piezo1-dependent calcium influx could modulate multiple down-stream signaling pathways and influence vascular remodelingthrough pleiotropic mechanisms. Future studies with conditionalknockouts will address whether endothelium and/or other celltypes are dependent on Piezo1-signaling in the embryo and inthe mature cardiovascular system.

Materials and MethodsAll animal procedures were approved by The Scripps Research Institute’sInstitutional Animal Care and Use Committee. Detailed methods are pro-vided in SI Materials and Methods.

X-gal and PLAP Stainings. Stainings were performed following standardprocedures. For X-gal staining, embryos were washed in X-gal washing buffer(PBS with 2 mM MgCl2, 0.02% Nonidet P-40, and 0.01% sodium deoxy-cholate), incubated in X-gal staining buffer (X-gal washing buffer with 5 mMpotassium ferricyanide, 5 mM potassium ferrocyanide, and 1 mg/mL X-gal)overnight with gentle shaking at 37 °C, and then postfixed in neutralbuffered 10% (vol/vol) formalin. Whole-mount embryos and tissues la-beled with X-gal were cleared in a 1:1 mixture of benzyl alcohol:benzylbenzoate clearing and imaged under Nikon SMZ1500 stereomicroscope (51).Images were obtained under Olympus AX70 microscope.

For PLAP staining, fixed embryos were heat-inactivated in PBS for 1.5 h at65 °C and washed in buffer B1 [0.15 M NaCl and 0.1 M Tris·HCl (pH7.5)] andbuffer B3 [0.1 M NaCl, 50 mMMgCl2, and 0.1 M Tris·HCl (pH 9.5)] sequentially.

Embryos were then incubated in NBT/BCIP solution (Roche) at room temper-ature overnight and postfixed. Images were obtained under Nikon SMZ1500stereomicroscope.

Flow Cytometry Analysis. Hearts and yolk sacs were isolated from embryosgenerated from timed matings between Piezo1+/P1-tdT mice. Yolk sacs andhearts were minced and incubated with 0.5 mg/mL collagenase type IV,0.1 mg/mL DNaseI, and 1 mg/mL trypsin inhibitor at 37 °C for 30 min, beingpipetted repeatedly several times throughout the incubation. The sampleswere washed with FACS buffer [PBS containing 2% (vol/vol) FBS and 2 mMEDTA] and centrifuged at 500 × g for 5 min. Erythrocytes were then lysedusing a multispecies RBC lysis buffer (eBioscience) for 5 min at roomtemperature, then washed again with FACS buffer. The cells were thenresuspended in FACS buffer, filtered through 100-μm-diameter nylonmesh, and incubated with FcBlock (BD) for 5 min at room temperature.They were then stained with an APC anti-PECAM1 antibody (BD) for15 min at 4 °C in the dark. The cells were washed once then resuspended inFACS buffer for data acquisition using a BD LSRII flow cytometer. Datawere analyzed using FlowJo (TreeStar).

LS Stimulation. Shear stress was applied using a perfusion tube (borosilicateglass capillary, pulled to an inner diameter 17 μm at the mouth) that wasfilled with extracellular solution (same as bath solution) and was positionedat an angle of 30° (sensapex SMX-series micromanipulation system; WarnerInstruments) relative to the cell. The micromanipulator was used to controlthe position of the perfusion tube and maintain consistency betweeneach trial. A Clampex 10.2 controlled pressure clamp HSPC-1 devise (ALAScientific) was connected to the perfusion tube and was used to applypressure on the solution to generate 600-ms pulses of flow. Starting from0 mmHg, increasing increments of 20 mmHg pressure was applied every 20 s

Fig. 5. Knockdown of PIEZO1 leads to defects in cellular alignment upon laminar shear stress. (A) F-actin images stained with TRITC-phalloidin (Upper) andangles of stress fiber orientation (Lower). Angle measurements were assigned a color code (right) where 0° is parallel to the flow chamber and −90° and 90°are perpendicular. (B) Histograms of stress fiber orientation of cells. (C) Histograms of cell orientation. The mean ± SEM values refer to percentage of ori-entations between ±15° for three independent biological experiments. Total numbers of cells analyzed are 348 (ST) and 236 (LS) for scrambled siRNA and 290(ST) and 170 (LS) for PIEZO1 siRNA. ST, static; LS, laminar shear for 24 h. Asterisk indicates significant difference by Watson’s U2 test.

Ranade et al. PNAS | July 15, 2014 | vol. 111 | no. 28 | 10351

PHYS

IOLO

GY

Dow

nloa

ded

by g

uest

on

July

25,

202

0

to raise the flow rate with each progressive pulse. Once a whole-cell inwardcurrent response was detected at a certain pressure level, the incrementswere reduced to 2 mmHg and the protocol was run again, starting from thelowest stimulus that elicited the initial response, to obtain a more accurateand sensitive stimulus–response curve for each cell. This data were then usedto obtain activation range and half-maximal activation values.

Laminar Flow Experiments for Cellular Alignment. HUVECs were seeded onglass slides coated with 50 μg/mL collagen-1 1 d before the experiment. Theslide was assembled into the flow chamber and exposed to laminar shear ata wall shear stress of 12 dynes/cm2. The cells obtained under static condition

were used as concurrent control groups. During the flow experiment, thesystem was kept at 37 °C and pH 7.5.

ACKNOWLEDGMENTS. We thank Taryn Goode for embryo analysis,Sergey Kupriyanov and Greg Martin at The Scripps Research InstituteMouse Genetics Core for embryonic stem cell injections, James Watsonand Terri Johnson for histology support, Kathy Spencer for help withimaging, and Dr. Hugh Rosen for mCherry antibody. This work wassupported by National Institutes of Health Grants R01 DE022358 (toA.P.) and 1R01HL121365 (to S.C.). S.S.R. is the recipient of a predoctoralfellowship from the California Institute for Regenerate Medicine. B.C. is therecipient of a postdoctoral fellowship from Dorris Neuroscience Center.

1. Hoffman BD, Grashoff C, Schwartz MA (2011) Dynamic molecular processes mediatecellular mechanotransduction. Nature 475(7356):316–323.

2. Gillespie PG, Müller U (2009) Mechanotransduction by hair cells: Models, molecules,and mechanisms. Cell 139(1):33–44.

3. Tsunozaki M, Bautista DM (2009) Mammalian somatosensory mechanotransduction.Curr Opin Neurobiol 19(4):362–369.

4. Orr AW, Helmke BP, Blackman BR, Schwartz MA (2006) Mechanisms of mechano-transduction. Dev Cell 10(1):11–20.

5. Jaalouk DE, Lammerding J (2009) Mechanotransduction gone awry. Nat Rev Mol CellBiol 10(1):63–73.

6. Hahn C, Schwartz MA (2009) Mechanotransduction in vascular physiology and ath-erogenesis. Nat Rev Mol Cell Biol 10(1):53–62.

7. Chalfie M (2009) Neurosensory mechanotransduction. Nat Rev Mol Cell Biol 10(1):44–52.

8. Corey DP, Hudspeth AJ (1979) Response latency of vertebrate hair cells. Biophys J26(3):499–506.

9. Guharay F, Sachs F (1984) Stretch-activated single ion channel currents in tissue-cultured embryonic chick skeletal muscle. J Physiol 352:685–701.

10. Delmas P, Hao J, Rodat-Despoix L (2011) Molecular mechanisms of mechano-transduction in mammalian sensory neurons. Nat Rev Neurosci 12(3):139–153.

11. Nilius B, Honoré E (2012) Sensing pressure with ion channels. Trends Neurosci 35(8):477–486.

12. Arnadottir J, Chalfie M (2010) Eukaryotic mechanosensitive channels. Annu Rev Bio-phys 39:111–137.

13. Geffeney SL, Goodman MB (2012) How we feel: Ion channel partnerships that detectmechanical inputs and give rise to touch and pain perception. Neuron 74(4):609–619.

14. Yan Z, et al. (2013) Drosophila NOMPC is a mechanotransduction channel subunit forgentle-touch sensation. Nature 493(7431):221–225.

15. Walker RG, Willingham AT, Zuker CS (2000) A Drosophila mechanosensory trans-duction channel. Science 287(5461):2229–2234.

16. O’Hagan R, Chalfie M, Goodman MB (2005) The MEC-4 DEG/ENaC channel of Cae-norhabditis elegans touch receptor neurons transduces mechanical signals. NatNeurosci 8(1):43–50.

17. Dedman A, et al. (2009) The mechano-gated K(2P) channel TREK-1. Eur Biophys J38(3):293–303.

18. Coste B, et al. (2010) Piezo1 and Piezo2 are essential components of distinct me-chanically activated cation channels. Science 330(6000):55–60.

19. Coste B, et al. (2012) Piezo proteins are pore-forming subunits of mechanically acti-vated channels. Nature 483(7388):176–181.

20. Kim SE, Coste B, Chadha A, Cook B, Patapoutian A (2012) The role of Drosophila Piezoin mechanical nociception. Nature 483(7388):209–212.

21. Woo SH, et al. (2014) Piezo2 is required for Merkel-cell mechanotransduction. Nature509(7502):622–626.

22. Maksimovic S, et al. (2014) Epidermal Merkel cells are mechanosensory cells that tunemammalian touch receptors. Nature 509(7502):617–621.

23. Ikeda R, et al. (2014) Merkel cells transduce and encode tactile stimuli to drive aβ-afferent impulses. Cell 157(3):664–675.

24. Albuisson J, et al. (2013) Dehydrated hereditary stomatocytosis linked to gain-of-function mutations in mechanically activated PIEZO1 ion channels. Nat Commun 4:1884.

25. Gallagher PG (2013) Disorders of red cell volume regulation. Curr Opin Hematol 20(3):201–207.

26. Coste B, et al. (2013) Gain-of-function mutations in the mechanically activated ionchannel PIEZO2 cause a subtype of Distal Arthrogryposis. Proc Natl Acad Sci USA110(12):4667–4672.

27. Leighton PA, et al. (2001) Defining brain wiring patterns and mechanisms through

gene trapping in mice. Nature 410(6825):174–179.28. Conway SJ, Kruzynska-Frejtag A, Kneer PL, Machnicki M, Koushik SV (2003) What

cardiovascular defect does my prenatal mouse mutant have, and why? Genesis 35(1):

1–21.29. Faucherre A, Nargeot J, Mangoni ME, Jopling C (2013) Piezo2b regulates vertebrate

light touch response. J Neurosci 33(43):17089–17094.30. Hove JR, et al. (2003) Intracardiac fluid forces are an essential epigenetic factor for

embryonic cardiogenesis. Nature 421(6919):172–177.31. Santhanakrishnan A, Miller LA (2011) Fluid dynamics of heart development. Cell Bi-

ochem Biophys 61(1):1–22.32. McGrath KE, Koniski AD, Malik J, Palis J (2003) Circulation is established in a stepwise

pattern in the mammalian embryo. Blood 101(5):1669–1676.33. le Noble F, et al. (2004) Flow regulates arterial-venous differentiation in the chick

embryo yolk sac. Development 131(2):361–375.34. Walls JR, Coultas L, Rossant J, Henkelman RM (2008) Three-dimensional analysis of

vascular development in the mouse embryo. PLoS ONE 3(8):e2853.35. Chien S (2007) Mechanotransduction and endothelial cell homeostasis: The wisdom

of the cell. Am J Physiol Heart Circ Physiol 292(3):H1209–H1224.36. Culver JC, Dickinson ME (2010) The effects of hemodynamic force on embryonic de-

velopment. Microcirculation 17(3):164–178.37. Willert CE, Gharib M (1991) Digital particle image velocimetry. Exp Fluids 10(4):

181–193.38. Peyronnet R, et al. (2013) Piezo1-dependent stretch-activated channels are inhibited

by Polycystin-2 in renal tubular epithelial cells. EMBO Rep 14(12):1143–1148.39. Davies PF (1995) Flow-mediated endothelial mechanotransduction. Physiol Rev 75(3):

519–560.40. Hahn C, Wang C, Orr AW, Coon BG, Schwartz MA (2011) JNK2 promotes endothelial

cell alignment under flow. PLoS ONE 6(8):e24338.41. Huang C, et al. (2003) Embryonic atrial function is essential for mouse embryogenesis,

cardiac morphogenesis and angiogenesis. Development 130(24):6111–6119.42. Yashiro K, Shiratori H, Hamada H (2007) Haemodynamics determined by a genetic

programme govern asymmetric development of the aortic arch. Nature 450(7167):

285–288.43. Lucitti JL, et al. (2007) Vascular remodeling of the mouse yolk sac requires hemody-

namic force. Development 134(18):3317–3326.44. O’Neil RG, Heller S (2005) The mechanosensitive nature of TRPV channels. Pflugers

Arch 451(1):193–203.45. Yamamoto K, et al. (2006) Impaired flow-dependent control of vascular tone and

remodeling in P2X4-deficient mice. Nat Med 12(1):133–137.46. Thi MM, Tarbell JM, Weinbaum S, Spray DC (2004) The role of the glycocalyx in re-

organization of the actin cytoskeleton under fluid shear stress: A “bumper-car”

model. Proc Natl Acad Sci USA 101(47):16483–16488.47. Mammoto A, Mammoto T, Ingber DE (2008) Rho signaling and mechanical control of

vascular development. Curr Opin Hematol 15(3):228–234.48. AbouAlaiwi WA, et al. (2009) Ciliary polycystin-2 is a mechanosensitive calcium

channel involved in nitric oxide signaling cascades. Circ Res 104(7):860–869.49. Tzima E, et al. (2005) A mechanosensory complex that mediates the endothelial cell

response to fluid shear stress. Nature 437(7057):426–431.50. Patel AJ, et al. (1998) A mammalian two pore domain mechano-gated S-like K+

channel. EMBO J 17(15):4283–4290.51. Schatz O, Golenser E, Ben-Arie N (2005) Clearing and photography of whole mount

X-gal stained mouse embryos. Biotechniques 39(5):650, 652, 654 passim.

10352 | www.pnas.org/cgi/doi/10.1073/pnas.1409233111 Ranade et al.

Dow

nloa

ded

by g

uest

on

July

25,

202

0